Using Machine Learning to Uncover Hidden Workforce Insights

Talent retention is Job #1 for most employers these days. But 69% of organizations believe they are failing in their efforts, Brandon Hall Group research shows, and a major reason is not understanding what employees expect from their employment. Only 38% of employers believe they have a strong or excellent understanding.

The irony is that most organizations are sitting on a goldmine of employee data that can drive better talent decisions. Technology has made it easy to collect and get access to the data. The problem is that it’s often siloed and disconnected from other data, making it difficult to use people data to drive decisions that can improve employee experience.

A second challenge is the growing amount of unstructured free-text data, which can be valuable for analyzing sentiment. Unstructured data is difficult and time-consuming to review — especially on a large scale, said Michael Cohen of Explorance, a Brandon Hall Group Platinum Preferred Provider. He presented at Brandon Hall Group’s HCM Excellence Conference 2023 in West Palm Beach, FL. 

In his session, Cohen explained how the Explorance BlueML comment analysis solution helps organizations interpret unstructured data and convert it into a structure that enables meaningful insights, analysis and strategic action.

Cohen said that machine learning works through a process called annotation, which converts unstructured data into structured data. Through annotation:

  • The comment or statement is analyzed and categorized into themes and subthemes. Examples might include workload, environment, knowledge, direct management and learning delivery.
  • Sentiments are assigned to each theme and subtheme. For example, does the comment convey a feeling that is positive, neutral or negative? 
  • Then the process uncovers any recommendations embedded in the comments, such as do less, do more, start, stop or change.

Through Machine Learning, Cohen said, this process occurs quickly and can be repeated hundreds of thousands of times to uncover hidden insights from unstructured data. Below is an illustration of how one comment (in the blue shaded area) is annotated automatically in Explorance BlueML.

Explorance’s research shows that the time to read a comment, assign two to three categories, identify sentiment and any recommendations can take between 15 and 25 seconds per comment. Doing this across a dataset of 5,000 comments would take between 21 and 35 hours. Explorance BlueML does it in a matter of minutes.

The tool gives employers the ability to leverage insights across multiple areas of HR — from performance reviews and engagement surveys to learning platforms and external sites such as Glassdoor, Indeed or Google. This can happen even if the data is not collected through Explorance, Cohen said, because the Explorance BlueML tool is technology agnostic.

Brandon Hall Group research shows that 60% of organizations are still pulling insights from annual engagement surveys — and often not much else — when all of this data, both structured and unstructured and constantly accumulating, can provide far richer and more detailed insights.

This type of technology can fundamentally change how organizations assess and address employee experience and can have a tremendous positive impact on talent retention. 

Click here for more information.

Like what you see? Share with a friend.

Related Content